English

• Energy consumption and environmental pollution have undoubtedly become two very important problems to be solved in today's world [1–3]. The development and utilization of solar energy holds great potential across various sectors [4–6]. Photocatalytic technology, as an environmentally friendly innovation, offers promising applications in energy production and environmental sustainability [7–9]. Among these, photocatalytic hydrogen production is a particularly viable strategy to reduce reliance on fossil fuels and promote clean energy initiatives [10–13]. Numerous photocatalysts have demonstrated effectiveness in hydrogen evolution [14–17]. In particular, the classical transition metal sulfide (CdS) is well-suited for this purpose owing to its narrow bandgap (2.4 eV), strong visible-light response, and high reduction capability [18–20]. Nevertheless, pristine CdS suffers from several intrinsic limitations, including low specific surface area, a scarcity of catalytically active sites, high photochemical corrosion, and rapid photogenerated charge recombination [21–23]. To address these issues, various strategies have been developed, such as metallic and nonmetallic doping, heterostructure construction and co-catalyst incorporation [24].

  Covalent organic frameworks (COFs) are crystalline polymers linked by strong covalent bonds, offering high specific surface area, abundant active sites, tunable structures, and excellent chemical stability [25–27]. Their adjustable band gaps and strong visible-light response make them promising photocatalysts [28,29]. Compared with early boroxyl- or borate-linked COFs, COFs based on imine, hydrazone, triazine, phenazine, or azide linkages are more stable and perform better in harsh environments [30]. It is well known that composite photocatalysts offer significant advantages by simultaneously addressing light absorption and redox potential. COFs can form various types of heterostructures with CdS. Not only do they possess matched interleaved band structures that allow active electrons and holes to accumulate in their respective bands with stronger redox capabilities, but the π-conjugated 2D crystalline structure of COFs compensates for CdS's poor stability and severe photocorrosion defects while enhancing its carrier mobility. Therefore, coupling CdS with COFs to form a heterojunction enables rapid carrier dynamics and strong interfacial contact, thereby effectively enhancing photocatalytic performance [31]. To date, such structures primarily consist of hollow spheres, core-shell structures and 0D/2D configurations [32,33]. Notably, Gao et al. and Feng et al. reported 2D/2D organic/inorganic direct Z-scheme TpTAP/CdS heterojunctions and CdS@COF core-shell heterojunctions, which demonstrate excellent hydrogen evolution performance [34,35]. Despite these advances, most reported COF/CdS structures still face practical challenges, such as reliance on toxic organic solvents, high energy consumption, and elevated reagent costs [36–38]. Consequently, developing COF/CdS structures that are facile to synthesize, energy-efficient, and environmentally benign remains a critical goal.

  In this work, a stable heterostructure was constructed between Cd_0.95Ni_0.05S (CNS) and a Schiff-based COF. By optimizing the COF content, the 15COF/CNS catalyst achieved a hydrogen production rate of 750.7 μmol h^-1 g^-1, which is 2.5 times higher than that of pristine CNS. The heterojunction effectively suppressed CNS photocorrosion and minimized photogenerated carrier recombination, while the porous COF structure facilitated proton exchange and hydrogen release. This work provides deep insight into the critical role of carrier dynamics in S-scheme heterojunctions and demonstrates substantial improvements in photocatalytic performance and hydrogen evolution efficiency.

  Fig. 1a illustrates the synthetic process of the COF/CNS heterostructure. The microstructure and morphology of COF/CNS were examined using transmission electron microscopy (TEM) and high-resolution TEM (HRTEM). As shown in Fig. 1b, amounts of nanoparticles are aggregated on the surface of the irregular bulk material. Combined with the corresponding energy dispersive X-ray spectroscopy (EDS) mapping results, the surface-aggregated particles are identified as CNS, while the bulk material corresponds to COF. This assignment is further verified by HRTEM, where a lattice spacing of 0.315 nm corresponding to the (101) crystalline plane of CNS, and lattice fringes with a spacing of 0.350 nm, attributed to the (100) plane of COF (Fig. 1c). Additionally, elemental mapping demonstrates a uniform distribution of C, N, O, Ni, S, and Cd in COF/CNS composite (Fig. 1d and Fig. S1 in Supporting information). Additionally, the actual doping amount of Ni is about 4.8 mol% (Table S1 in Supporting information), which closely aligns with the theoretical doping ratio, thereby verifying the accuracy of the doping content. These results indicate the successful preparation of COF/CNS composites.

  Figure 1

  

  Figure 1.  (a) Schematic illustration of the synthetic process of the COF/CNS heterostructure. TEM (b), HRTEM (c), and elemental mapping images (d) of 15COF/CNS.

  DownLoad: Full-Size Img PowerPoint

  The crystalline structures of materials were analyzed using X-ray diffraction (XRD) (Figs. 2a and b, and Fig. S2 in Supporting information). The pristine CNS exhibits a hexagonal zinc blende phase, with characteristic diffraction peaks at 24.8°, 26.5°, 28.2°, 36.6°, 43.7°, and 47.8° corresponding to the (100), (002), (101), (102), (110), and (103) crystal planes, respectively, in agreement with JCPDS card #41–1049 [39]. The XRD pattern of COF shows peaks at 2.8°, 4.8°, 5.6°, 7.4°, 9.7°, and 25.4°, corresponding to the (100), (110), (200), (210), (220), and (001) crystal planes, respectively [40]. Notably, COF/CNS is basically reminiscent of pristine CNS in terms of the XRD patterns. In addition, no other peaks were detected in the samples, suggesting that no other crystalline phases were generated during synthesis.

  Figure 2

  

  Figure 2.  XRD (a), small-angle XRD (b), FTIR (c), and N₂ adsorption–desorption (d) of COF, CNS, and 15COF/CNS. High resolution XPS spectra of C 1s (e) and N 1s (f) for COF and 15COF/CNS. High resolution XPS spectra of S 2p (g), Cd 3d (h), Ni 2p (i) for CNS and 15COF/CNS.

  DownLoad: Full-Size Img PowerPoint

  Solid-state ¹³C NMR further validated the chemical structure of COFs, with signals at δ 55.24 (methoxy groups), δ 110–154 (benzene rings), δ 189.03 and 207.06 (C-structures) aligning with theoretical predictions (Fig. S3 in Supporting information) [41]. The Fourier transform infrared (FTIR) spectra of CNS, COF, and COF/CNS were measured (Fig. 2c and Fig. S4 in Supporting information). In the spectrum of CNS, the stretching vibration peaks of the Cd-S group appear at 685 cm⁻¹ [42]. When COF was grown on the surface of –NH₂ modifed CNS to form the COF/CNS structure, a characteristic imine bond stretching vibration band was also observed at 1605 cm⁻¹, suggesting the successful recombination of CNS and COF [43].

  The pore structures of catalysts were investigated by nitrogen adsorption–desorption measurements (Fig. 2d and Table S2 in Supporting information). The bare COF possesses a type IV isotherm according to the IUPAC classification with a high Brunauer-Emmett-Teller (BET) surface area of 2069.8 m²/g [44]. Notably, the surface area of 15COF/CNS is about 7 times higher than that of CNS, owing to the incorporation of COF. All synthesized samples exhibit mesoporous structures (Fig. S5 in Supporting information). This mesoporous structure and large specific surface area are conducive to the progression of photocatalytic reactions. The incorporation of COF enhances the specific surface area of the composite, thereby providing more active sites for photocatalytic reactions.

  The X-ray photoelectron spectroscopy (XPS) spectrum of 15COF/CNS confirms the presence of C, O, N, Cd, S, and Ni elements (Fig. S6 in Supporting information). The high-resolution C 1s spectrum of COF (Fig. 2e) exhibits peaks at 284.80, 286.20, and 288.89 eV, attributed to aromatic sp² carbon, C—C and C—N bonds, respectively [45]. The N 1s spectrum of COF (Fig. 2f) shows peaks at 398.72, 399.44, and 400.60 eV, attributed to C–N and N–H groups [46]. For 15COF/CNS, the C 1s and N 1s shift to lower binding energy, indicating obtain electrons. The S 2p spectrum of CNS (Fig. 2g) shows peaks at 161.26 and 162.19 eV, corresponding to S^2- (S 2p_3/2 and S 2p_1/2) [47]. The Cd 3d spectrum of CNS (Fig. 2h) presents peaks at 405.08 and 411.79 eV, characteristic of Cd²⁺ (Cd 3d_5/2 and Cd 3d_3/2) [48]. The Ni 2p spectrum (Fig. 2i) shows peaks at 854.72 and 878.47 eV, indicative of Ni²⁺ (Ni 2p_3/2 and Ni 2p_1/2), along with additional peaks at 860.51 and 873.01 eV, corresponding to Ni³⁺ [49]. For 15COF/CNS, the S 2p, Cd 3d, and Ni 2p shift to higher binding energy, indicating a loss of electrons.

  The photocatalytic hydrogen evolution performance of catalysts was evaluated under light irradiation using sodium ascorbate (NaAsc) as a hole-trapping sacrificial agent and trace amounts of Co²⁺ as a cocatalyst. Nickel-doped CdS exhibits enhanced hydrogen production rates, ultraviolet absorption, and photocurrent, along with a reduced impedance arc radius, confirming that Ni doping effectively mitigates the photocorrosion of CdS (Fig. S7 in Supporting information). The photocatalytic performance of Ni-doped and composite COF materials (COF/CNS) is further enhanced. As shown in Fig. 3a, the pristine COF exhibits negligible photocatalytic activity, whereas pristine CNS achieves a H₂ evolution rate of 305.2 µmol h^-1 g^-1, which remains insufficient for practical applications. In contrast, the COF/CNS composites display markedly enhanced H₂ production rates. With increasing the mass ratio of COF to CNS, the H₂ evolution efficiency of COF/CNS first improves and then declines. The H₂ production rate reaches a maximum value of 750.7 µmol h^-1 g^-1 for 15COF/CNS, which is approximately 2.5 times and 3753.5 times that of CNS and COF, respectively. Fig. 3b demonstrates the superiority of NaAsc as a sacrificial agent. Compared to other sacrificial agents, NaAsc exhibits a redox potential more compatible with COF/CNS, thereby lowering the reaction energy barrier. It maintains good activity under weakly acidic conditions, facilitating the reduction of protons by photoelectrons to generate hydrogen [50]. Furthermore, NaAsc itself is non-toxic and non-polluting, making it environmentally friendly. Without the addition of Co²⁺, the hydrogen production rate decreases significantly, as shown in Fig. 3c. This is because the addition of trace amounts of Co²⁺ reduces the overpotential of the hydrogen evolution reaction, thereby enhancing the reaction rate [51]. Co²⁺ consumes accumulated electrons, mitigates the photocorrosion of CdS, and improves catalyst stability. Additionally, it reduces the amount of precious metals required, lowering production costs. Figs. 3d and e demonstrate that combined illumination with visible and ultraviolet light enhances hydrogen production performance, and that hydrogen generation is further favored under weakly acidic pH conditions. Under these pH conditions, the H⁺ concentration is optimal, promoting electron migration. After 3 cycles with a time interval of 1 day, the COF/CNS composite maintained a high activity of H₂ generation (Fig. 3f), which is quite close to the initial performance if the mass loss was taken into account. Additionally, the crystal structure (Fig. S8 in Supporting information) of COF/CNS after cycling tests maintained the same as the original sample, demonstrating its outstanding stability. These results highlight the critical role of CNS-based heterostructures in boosting photocatalytic H₂ production and underscore the favorable influence of heterojunctions on photocatalytic performance (Fig. S9 and Table S3 in Supporting information).

  Figure 3

  

  Figure 3.  (a) H₂ evolution rates of CNS, COF, and COF/CNS composites with different mass ratios. The effects of sacrificial agent type (b), Co²⁺ cocatalyst (c), light exposure (d), and buffer solution pH (e) on H₂ evolution rate. (f) H₂ evolution in the cycle test for the COF/CNS catalyst. (g) Tauc plots of CNS, COF, and 15COF/CNS. Mott-Schottky curves of CNS (h) and COF (i).

  DownLoad: Full-Size Img PowerPoint

  The bandgaps of catalysts were determined using the Kubelka-Munk equation: (αhv)ⁿ = K (hv - E_g) [52]. The Tauc plot (Fig. 3g) reveals that the direct bandgaps (E_g) of CNS, COF, and 15COF/CNS are 2.23, 2.44, and 2.17 eV, respectively. The flat band potentials of the photocatalysts were assessed by Mott-Schottky measurements (Figs. 3h and i). The positive slopes of the curves confirm the n-type semiconductor behavior of all catalysts [53,54]. The conduction band (CB) potentials of CNS and COF are −1.01 and −0.62 V (vs. Ag/AgCl), corresponding to −0.81 and −0.42 V after conversion to the normal hydrogen electrode (NHE) reference electrode. Based on the relationship E_VB = E_CB + E_g [55], the VB potentials of CNS and COF are calculated as 1.42 and 2.02 V, respectively (Fig. S10 in Supporting information). This energy band alignment indicates that COF and CNS can form an S-scheme heterojunction, which is favorable for efficient charge separation and photocatalytic performance.

  The photocurrent measurements were conducted to evaluate the efficiency of photogenerated carrier separation and transfer. As shown in Fig. 4a, 15COF/CNS exhibits the highest photocurrent density among all tested photocatalysts, indicating that COF incorporation significantly promotes the separation of photoexcited electrons and holes [56]. The Nyquist plots from electrochemical impedance spectroscopy (EIS) (Fig. 4b) reveal a semicircle diameter order of COF > CNS > 15COF/CNS. The smaller semicircle diameter for 15COF/CNS indicates enhanced interfacial charge transfer, which contributes to its superior photocatalytic performance. Further analysis of the separation and transfer of photogenerated carriers was performed using the time-resolved photoluminescence (TRPL) spectroscopy (Fig. S11 and Table S4 in Supporting information). Compared with pure CNS (4.28 ns) and COF (4.39 ns), the COF/CNS composite (6.26 ns) exhibited a longer carrier lifetime, suggesting efficient carrier transport performance [57]. The optical absorption properties were evaluated using UV–visible diffuse reflectance spectroscopy (UV-vis DRS). As shown in Fig. 4c, CNS displays an optical absorption edge at approximately 540 nm, which is larger than that pristine COF (500 nm) and CNS (530 nm). Based on the above results, it concludes that the constructed COF/CNS heterojunction can highly improve the charge separation efficiency and reduce the migration resistance of electrons and holes, which is beneficial for improving the photocatalytic H₂ evolution.

  Figure 4

  

  Figure 4.  Photocurrent response plots (a), EIS Nyquist plots (b), and UV–vis DRS (c) of CNS, COF and 15COF/CNS. fs-TA spectra of CNS (d) and 15COF/CNS (g) with delay time from 8 ps to 3.3 ns, pseudocolor fs-TA plots of CNS (e) and 15COF/CNS (h). Normalized ESA recovery kinetics of CNS (f) probed at 460 nm and 15COF/CNS (i) probed at 470 nm with a delay time within 2000 ps.

  DownLoad: Full-Size Img PowerPoint

  To further elucidate the charge carrier dynamics, femtosecond transient absorption (fs-TA) spectroscopy was employed. The TA spectra of CNS and 15COF/CNS (Figs. 4d and g) exhibit distinct peaks near 460 and 470 nm, respectively, corresponding to excited-state relaxation. Compared with CNS, 15COF/CNS exhibits a faster decay rate, as evidenced by the pseudo-colored fs-TA maps (Figs. 4e and h) and time-resolved profiles. The fs-TA decay kinetics were fitted at 460 nm. As shown in Fig. 4f, the faster decay time constant of τ₁ (214.90 ps) in CNS is due to the trapping of photogenerated electrons by trap states near the CB edge of CNS, and the slower decay time constant of τ₂ (590.77 ps) is attributed to the recombination between electrons of CB and the holes of VB. Compared with CNS, the reduced decay time constants of τ₁ (117.17 ps) and τ₂ (260.85 ps) in 15COF/CNS suggest that the electron relaxation is greatly accelerated (Fig. 4i), which is ascribed to the directional migration of charge carriers retarding their reverse transport. In addition, the decay time constants of 15COF/CNS are evidently reduced as illustrated in Fig. 4i, thereby proving that more efficient charge transport occurs in 15COF/CNS. The ultra-fast delay time (τ₃ = 10.46 ps) corresponds to an additional dynamic process, which is attributed to charge transfer at the COF/CNS interface [58,59].

  The interfacial charge transfer mechanism in the catalyst was investigated using photo-activated Kelvin probe force microscopy (KPFM). As shown in Figs. 5a-d, the surface potential distributions of CNS and 15COF/CNS were measured before and after illumination. The average surface potential of the CNS increased from 118 mV in the dark to 153 mV after irradiation, representing a potential difference of approximately 35 mV. In contrast, the surface potential of 15COF/CNS rose from −5 mV to 79 mV, exhibiting a potential difference of 84 mV (Figs. 5e and f). The results indicate that the potential difference of 15COF/CNS is significantly enhanced compared to CNS. This is attributed to light-driven electron migration from COF to CNS. This electron transfer forms an internal electric field (IEF) at its contact interface to accelerate the separation of electrons and holes and enhance transfer efficiency [60].

  Figure 5

  

  Figure 5.  KPFM images of CNS (a, b) and 15COF/CNS with or without light irradiation (c, d). The line-scanning surface potential of CNS (e) and 15COF/CNS (f). In situ C 1s (g), N 1s (h), Cd 3d (i), and S 2p (j) XPS spectra of 15COF/CNS. (k) Charge carrier transfer mechanism in the COF/CNS heterojunction.

  DownLoad: Full-Size Img PowerPoint

  To confirm the photoreactivity mechanism, we conducted the electron spin resonance (ESR) analysis (Fig. S12 in Supporting information) with 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as the trapping agent to detect superoxide radicals (·O₂^–) and hydroxyl radical (·OH) in an aqueous solution. No signals were observed in the dark for all the samples, indicating that ·O₂^– and ·OH generation requires light stimulation. This indicates that COF/CNS has both the more oxidative holes in COF and the more reductive electrons in CNS [61]. This observation indicates that the charge transfer path between COF and CNS aligns with S-scheme charge transfer.

  Moreover, in-situ irradiated XPS spectra were conducted to investigate the interfacial charge transfer between CNS and COF. In the dark, the C 1s and N 1s peaks of COF/CNS shift to lower binding energies compared with COF (Figs. 5g and h), while the Cd 3d, S 2p, and Ni 2p peaks shift to higher energies relative to CNS (Figs. 5i and j, and Fig. S13 in Supporting information). This suggests electron migration from CNS to COF during composite formation, generating an IEF directed from CNS to COF. Under illumination, the C 1s and N 1s peaks shift to higher energies, whereas Cd 3d, Ni 2p, and S 2p shift to lower energies. These results indicate electron transfer from CNS to COF upon photoexcitation, providing strong evidence for an S-scheme charge migration pathway at the COF/CNS interface. DFT calculations show the charge distribution and confirmed that 0.89 electrons were transferred from CNS to COF, which is consistent with the XPS test results (Fig. S14 in Supporting information). Ultraviolet photoelectron spectroscopy (UPS) show that the work functions (W_f) of CNS and COF are 3.35 and 3.52 eV, respectively, indicating that the Fermi level of CNS is higher than that of COF (Fig. S15 in Supporting information). When the CNS/COF heterojunction forms, electrons tend to transfer from CNS to COF, which is consistent with the XPS results.

  The composite materials demonstrate enhanced light absorption capacity, prolonged lifetime of photogenerated carrier, and increased active sites, resulting in significantly improved hydrogen evolution activity compared to unmodified samples. The hydrogen evolution mechanism on COF-modified CNS is illustrated in Fig. 5k. Nickel doping introduces an intermediate bandgap state, altering the electronic band structure of CdS and enhancing its light absorption capacity. This intermediate state also prolongs the lifetime of photo-generated carriers. After COF modification, the number of active sites increases, facilitating proton adsorption and H₂ desorption. Under illumination, both COF and CNS are excited, with electrons being excited to their respective CB and holes being generated in the valence band. Electrons in the CB of COF transfer to the VB of CNS and recombine with holes, while excess electrons in the CB of CNS participate in hydrogen generation. Holes in the VB of COF drive the oxidation reaction. This S-scheme charge transfer mechanism accelerates charge separation, enhances activity, and promotes rapid hydrogen generation.

  In conclusion, a COF/CNS heterostructure composite was successfully synthesized and thoroughly characterized. The integration of COF endows the composite with a large surface area and tunable architectures, enabling efficient charge transfer and enhanced light absorption, thereby markedly improving photocatalytic hydrogen evolution. The optimized 15COF/CNS achieved a hydrogen production rate of 750.7 µmol h^-1 g^-1, 2.5 times higher than of pristine CNS. This enhancement is attributed to the S-scheme charge transfer mechanism, which promotes carrier separation, suppresses electron–hole recombination, and mitigates CNS photocorrosion. Moreover, COF incorporation further increases the specific surface area and strengthens light-harvesting capability. Overall, this study presents a promising strategy for developing cost-effective and efficient CdS-based photocatalysts.

  CRediT authorship contribution statement

  Yue Zhao: Data curation. Kezhen Qi: Writing – review & editing. Shan Wang: Formal analysis. Li Zhao: Formal analysis.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  Acknowledgments

  This research was financially supported by the National Natural Science Foundation of China (Nos. 22268003 and 22568002) and Yunnan Fundamental Research Projects (Nos. 202305AF150116, 202401BA070001-035 and 202505AF350065).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2026.112421.

  
  
• 1. [1]

     G. Cheng, X. Liu, J.Y. Xiong, et al., Chem. Eng. J. 501 (2024) 157491. doi: 10.1016/j.cej.2024.157491

  2. [2]

     L.Y. Zhu, X. Tian, G. Shi, et al., Chin. Chem. Lett. 36 (2025) 111088. doi: 10.1016/j.cclet.2025.111088

  3. [3]

     K.Z. Qi, Z.D. Wei, H.B. Wang, et al., Chin. Chem. Lett. 30 (2026) 111661.

  4. [4]

     M. Zhou, H.F. Wang, Chin. Chem. Lett. 33 (2022) 4705–4709. doi: 10.1016/j.cclet.2021.12.074

  5. [5]

     X.B. Li, Y.Y. Wan, F. Deng, et al., Chin. Chem. Lett. 36 (2025) 111418. doi: 10.1016/j.cclet.2025.111418

  6. [6]

     X.Y. Ye, Y.C. Dong, W.B. Hua, et al., Carbon Energy 5 (2025) e70024.

  7. [7]

     C. Guo, X.P. Tian, Z.X. Zhuang, et al., Chem. Eng. J. 521 (2025) 166469. doi: 10.1016/j.cej.2025.166469

  8. [8]

     R.J. Li, H.X. Li, X.D. Zhang, et al., Adv. Funct. Mater. 34 (2024) 2402797. doi: 10.1002/adfm.202402797

  9. [9]

     X.Y. Meng, S.Y. Wang, C.C. Zhang, et al., ACS Catal. 12 (2022) 10115–10126. doi: 10.1021/acscatal.2c01877

  10. [10]

      J.X. Bai, R.C. Shen, Z.M. Jiang, et al., Chin. J. Catal. 43 (2022) 359–369. doi: 10.1016/S1872-2067(21)63883-4

  11. [11]

      X.Q. Hao, Y.C. Wang, J. Zhou, et al., Appl. Catal. B: Environ. 221 (2018) 302–311. doi: 10.1016/j.apcatb.2017.09.006

  12. [12]

      Y. Wu, C. Cheng, K.Z. Qi, et al., Acta Phys. Chim. Sin. 40 (2024) 2406027. doi: 10.3866/pku.whxb202406027

  13. [13]

      J.J. Zhang, K.Z. Qi, R. Pitcheri, et al., J. Photochem. Photobiol. C Photochem. Rev. 65 (2025) 100722. doi: 10.1016/j.jphotochemrev.2025.100722

  14. [14]

      X.T. Liu, W. Li, Z. Liu, et al., Chin. Chem. Lett. 37 (2026) 111457. doi: 10.1016/j.cclet.2025.111457

  15. [15]

      P. Qiu, J.Y. Xiong, M.J. Lu, et al., J. Colloid Interf. Sci. 622 (2022) 924–937. doi: 10.1016/j.jcis.2022.04.107

  16. [16]

      C.Y. Wang, J.Y. Xiong, Z.P. Wen, et al., Ind. Eng. Chem. Res. 62 (2023) 11402–11413. doi: 10.1021/acs.iecr.3c01302

  17. [17]

      Y. Wei, G. Cheng, J.Y. Xiong, et al., J. Energy Chem. 32 (2019) 45–56. doi: 10.1117/12.2542125

  18. [18]

      Y. Tang, Y. Sun, Y.F. Li, et al., Adv. Funct. Mater. 34 (2024) 2405527. doi: 10.1002/adfm.202405527

  19. [19]

      Y.C. Zhang, X.L. Ran, H.T. Fu, et al., Adv. Funct. Mater. 34 (2024) 2404585. doi: 10.1002/adfm.202404585

  20. [20]

      Z. Qi, J.B. Chen, Q. Li, et al., Small 19 (2023) 2303318. doi: 10.1002/smll.202303318

  21. [21]

      H.S. Zhai, X.L. Liu, P. Wang, et al., Appl. Surf. Sci. 430 (2018) 515–522. doi: 10.1016/j.apsusc.2017.08.056

  22. [22]

      Y.Q. Li, S.J. Wan, W.C. Liang, et al., Small 20 (2024) 2312104. doi: 10.1002/smll.202312104

  23. [23]

      Z.Y. Xu, H.J. Liu, J.L. Yang, et al., Adv. Mater. 37 (2025) 2501091. doi: 10.1002/adma.202501091

  24. [24]

      W.X. Zhang, J.Y. Xiong, S.Z. Li, et al., Mol. Catal. 570 (2025) 114701.

  25. [25]

      H. Guo, Z.Y. Yang, L. Sun, et al., J. Colloid Interf. Sci. 659 (2024) 639–649. doi: 10.1016/j.jcis.2023.12.180

  26. [26]

      Y.L. Chen, Y. Xie, X. Sun, et al., Sens. Actuators B: Chem. 331 (2021) 129438. doi: 10.1016/j.snb.2021.129438

  27. [27]

      Y.R. Du, B.H. Xu, J.S. Pan, et al., Green Chem. 21 (2019) 4792–4799. doi: 10.1039/c9gc02126d

  28. [28]

      C.Q. Wang, S.H. Shi, F. Duan, et al., J. Mater. Chem. A 10 (2022) 16524–16532. doi: 10.1039/d2ta03461a

  29. [29]

      Y.L. Xu, X.F. Shi, R. Hua, et al., Appl. Catal. B: Environ. 260 (2020) 118142. doi: 10.1016/j.apcatb.2019.118142

  30. [30]

      J.X. Liu, K.Z. Qi, X.L. Xiang, et al., Energy Environ. Mater. 8 (2025) e70071. doi: 10.1002/eem2.70071

  31. [31]

      G.E. Tang, J.J. Zhang, C.B. Bie, et al., Adv. Mater. 37 (2025) e14576. doi: 10.1002/adma.202514576

  32. [32]

      K.Y. Zhang, G.L. Lu, Z.S. Xi, et al., Chin. Chem. Lett. 32 (2021) 2207–2211. doi: 10.1016/j.cclet.2020.12.021

  33. [33]

      Y.X. He, J.S. Zhao, Y. Sham, et al., ACS Sustain. Chem. Eng. 11 (2023) 17552–17563. doi: 10.1021/acssuschemeng.3c06421

  34. [34]

      R.Q. Gao, J.X. Bai, R.C. Shen, et al., J. Mater. Sci. Technol. 137 (2023) 223–231. doi: 10.1016/j.jmst.2022.09.001

  35. [35]

      Y.N. Feng, J.J. Li, S.H. Ye, et al., Sustain. Energy Fuels 6 (2022) 5089. doi: 10.1039/d2se01159j

  36. [36]

      C. Sun, L. Karuppasamy, L. Gurusamy, et al., Sep. Purif. Technol. 271 (2021) 118873. doi: 10.1016/j.seppur.2021.118873

  37. [37]

      D.D. Liu, C. Zhu, G. Yuan, et al., Mater. Lett. 365 (2024) 136417. doi: 10.1016/j.matlet.2024.136417

  38. [38]

      K.F. Yu, P. He, N.N. He, et al., Sci. China Mater. 66 (2023) 4680–4688. doi: 10.1007/s40843-023-2599-9

  39. [39]

      Q.W. Dong, M.L. Li, M.S. Sun, et al., Small Methods 5 (2021) e2100878. doi: 10.1002/smtd.202100878

  40. [40]

      X.F. Shi, Y.J. Yao, Y.L. Xu, et al., ACS Appl. Mater. Interfaces 9 (2017) 7481–7488. doi: 10.1021/acsami.6b16267

  41. [41]

      C.H. Li, W.Z. She, J.L. Liu, et al., Colloids Surf. A: Physicochem. Eng. Asp. 691 (2024) 133841. doi: 10.1016/j.colsurfa.2024.133841

  42. [42]

      G.T. Sun, J.J. Zhang, B. Cheng, et al., Chem. Eng. J. 476 (2023) 146818. doi: 10.1016/j.cej.2023.146818

  43. [43]

      T. Zhang, C.W. Gao, W. Huang, et al., Talanta 188 (2018) 578. doi: 10.1016/j.talanta.2018.06.032

  44. [44]

      H.W. Zhang, Q.Q. Zhu, R.R. Yuan, et al., Sens. Actuators B: Chem. 329 (2021) 129144. doi: 10.1016/j.snb.2020.129144

  45. [45]

      Y. Tang, W.Y. Li, Y. Muhammad, et al., Chem. Eng. J. 421 (2021) 129743. doi: 10.1016/j.cej.2021.129743

  46. [46]

      L. Sun, L.L. Li, J. Yang, et al., Chin. J. Catal. 43 (2022) 350–358. doi: 10.1016/S1872-2067(21)63869-X

  47. [47]

      P.L. Liang, L.Y. Yuan, K. Du, et al., Chem. Eng. J. 420 (2021) 129831. doi: 10.1016/j.cej.2021.129831

  48. [48]

      C. Zhang, Z.Z. Ai, X.L. Xu, et al., Small 21 (2025) 2411128.

  49. [49]

      N. Li, Y.P. Qiu, L.P. Li, et al., Small 21 (2025) 2408057. doi: 10.1002/smll.202408057

  50. [50]

      X.Y. Zhang, X. Wu, R.F. Chen, et al., J. Colloid Interf. Sci. 665 (2024) 100–108. doi: 10.1016/j.jcis.2024.03.104

  51. [51]

      R.M. Irfan, D.C. Jiang, Z.J. Sun, et al., J. Catal. 353 (2017) 274–285. doi: 10.1016/j.jcat.2017.06.010

  52. [52]

      Y.Y. Qin, H. Li, J. Lu, et al., Chem. Eng. J. 384 (2020) 123275. doi: 10.1016/j.cej.2019.123275

  53. [53]

      J.Y. Qiu, K. Meng, Y. Zhang, et al., Adv. Mater. 36 (2024) 2400288. doi: 10.1002/adma.202400288

  54. [54]

      W.Z. Li, Y. Liu, Z.Y. Wang, et al., J. Environ. Chem. Eng. 12 (2024) 114204. doi: 10.1016/j.jece.2024.114204

  55. [55]

      K.Y. Lin, L. Feng, D.K. Li, et al., Appl. Catal. B: Environ. 286 (2021) 119880. doi: 10.1016/j.apcatb.2021.119880

  56. [56]

      J. Wang, Y.Q. Zhang, S.J. Jiang, et al., Angew. Chem. Int. Ed. 62 (2023) e202307808. doi: 10.1002/anie.202307808

  57. [57]

      B. Qi, R.C. Shen, Z.Q. Ren, J. Mater. Sci. Technol. 232 (2025) 65–73. doi: 10.1016/j.jmst.2025.03.003

  58. [58]

      X. Fei, S. Zhang, D. Zhai, et al., Chem. Sci. 13 (2022) 9914–9920. doi: 10.1039/d2sc00263a

  59. [59]

      N. Lv, S. Lin, L. Ding, et al., Chem. Eng. J. 522 (2025) 167143. doi: 10.1016/j.cej.2025.167143

  60. [60]

      M. Wei, C. Cheng, B.W. He, et al., Acta Phys. Chim. Sin. 41 (2025) 100158. doi: 10.1016/j.actphy.2025.100158

  61. [61]

      J.Y. Lu, S.Y. Liu, R. Pitcheri, et al., Chem. Res. Chin. Univ. 41 (2025) 799–811. doi: 10.1007/s40242-025-5089-6

• 

  Figure 1  (a) Schematic illustration of the synthetic process of the COF/CNS heterostructure. TEM (b), HRTEM (c), and elemental mapping images (d) of 15COF/CNS.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  XRD (a), small-angle XRD (b), FTIR (c), and N₂ adsorption–desorption (d) of COF, CNS, and 15COF/CNS. High resolution XPS spectra of C 1s (e) and N 1s (f) for COF and 15COF/CNS. High resolution XPS spectra of S 2p (g), Cd 3d (h), Ni 2p (i) for CNS and 15COF/CNS.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) H₂ evolution rates of CNS, COF, and COF/CNS composites with different mass ratios. The effects of sacrificial agent type (b), Co²⁺ cocatalyst (c), light exposure (d), and buffer solution pH (e) on H₂ evolution rate. (f) H₂ evolution in the cycle test for the COF/CNS catalyst. (g) Tauc plots of CNS, COF, and 15COF/CNS. Mott-Schottky curves of CNS (h) and COF (i).

  下载: 全尺寸图片 幻灯片
  

  Figure 4  Photocurrent response plots (a), EIS Nyquist plots (b), and UV–vis DRS (c) of CNS, COF and 15COF/CNS. fs-TA spectra of CNS (d) and 15COF/CNS (g) with delay time from 8 ps to 3.3 ns, pseudocolor fs-TA plots of CNS (e) and 15COF/CNS (h). Normalized ESA recovery kinetics of CNS (f) probed at 460 nm and 15COF/CNS (i) probed at 470 nm with a delay time within 2000 ps.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  KPFM images of CNS (a, b) and 15COF/CNS with or without light irradiation (c, d). The line-scanning surface potential of CNS (e) and 15COF/CNS (f). In situ C 1s (g), N 1s (h), Cd 3d (i), and S 2p (j) XPS spectra of 15COF/CNS. (k) Charge carrier transfer mechanism in the COF/CNS heterojunction.

  下载: 全尺寸图片 幻灯片
•